Wnt signaling inhibits osteogenic differentiation of human mesenchymal stem cells

www.elsevier.com/locate/bone

Bone 34 (2004) 818–826

Wnt signaling inhibits osteogenic differentiation of human

mesenchymal stem cells

Jan de Boer,a,* Ramakrishnaiah Siddappa,a Claudia Gaspar,b,c Aart van Apeldoorn,a

Ricardo Fodde,b,c and Clemens van Blitterswijka

a Institute for Biomedical Technology, University of Twente, Enschede, The NetherlandsbCenter for Human and Clinical Genetics, Leiden University Medical Center, Leiden, The Netherlands

cDepartment of Pathology, Josephine Nefkens Institute, Erasmus Medical Center, Rotterdam, The Netherlands

Received 15 October 2003; revised 24 December 2003; accepted 22 January 2004

Abstract

Human mesenchymal stem cells (hMSCs) from the bone marrow represent a potential source of pluripotent cells for autologous bone tissue

engineering. We previously discovered that over activation of the Wnt signal transduction pathway by either lithium or Wnt3A stimulates

hMSC proliferation while retaining pluripotency. Release of Wnt3A or lithium from porous calcium phosphate scaffolds, which we use for

bone tissue engineering, could provide a mitogenic stimulus to implanted hMSCs. To define the proper release profile, we first assessed the

effect of Wnt over activation on osteogenic differentiation of hMSCs. Here, we report that both lithium and Wnt3A strongly inhibit

dexamethasone-induced expression of the osteogenic marker alkaline phosphatase (ALP). Moreover, lithium partly inhibited mineralization of

hMSCs whereas Wnt3A completely blocked it. Time course analysis during osteogenic differentiation revealed that 4 days of Wnt3A exposure

before the onset of mineralization is sufficient to block mineralization completely. Gene expression profiling in Wnt3A and lithium-exposed

hMSCs showed that many osteogenic and chondrogenic markers, normally expressed in proliferating hMSCs, are downregulated upon Wnt

stimulation. We conclude that Wnt signaling inhibits dexamethasone-induced osteogenesis in hMSCs. In future studies, we will try to limit

release of lithium or Wnt3A from calcium phosphate scaffolds to the proliferative phase of osteogenesis.

D 2004 Elsevier Inc. All rights reserved.

Keywords: Human mesenchymal stem cells; Wnt signaling; Osteogenesis; Tissue engineering; Micro-array

Introduction glycerol phosphate, hMSCs express osteogenic markers such

Human mesenchymal stem cells (hMSCs) are pluripotent

cells from the bone marrow, which can be expanded in vitro

and differentiated into the osteogenic, chondrogenic, and

adipogenic lineages [36]. MSCs were initially identified as

the fibroblastic adherent fraction of bone marrow aspirates

[6,16] and are also called colony forming units-fibroblasts

(CFU-F), marrow stromal cells, bone marrow mesenchymal

cells, or mesenchymal progenitor cells. In vitro osteogenic

differentiation of hMSCs recapitulates many of the develop-

mental steps during normal in vivo osteogenesis. For in-

stance, in the presence of dexamethasone (dex) and h-

8756-3282/$ - see front matter D 2004 Elsevier Inc. All rights reserved.

doi:10.1016/j.bone.2004.01.016

* Corresponding author. Institute for Biomedical Technology, Univer-

sity of Twente, Prof. Bronkhorstlaan 10D, 3723 MB, Bilthoven, The

Netherlands. Fax: +31-30-2280255.

E-mail address: [email protected] (J. de Boer).

as bone-specific alkaline phosphatase (ALP) and they de-

posit an extracellular matrix, which becomes mineralized

under appropriate culture conditions [5,8,22,32,37]. Because

of their ready availability and well-established in vitro

culturing protocols, hMSCs have been the source of cells

in autologous bone and cartilage tissue engineering

[2,7,18,33]. For bone tissue engineering, we and others have

demonstrated ectopic bone formation by seeding hMSCs

onto porous calcium phosphate scaffolds and subsequent

subcutaneous implantation into immune-deficient mice

[11,20].

To further improve bone tissue engineering protocols

using hMSCs, we are interested in molecular cues that can

stimulate hMSC proliferation and differentiation both in

vitro and in vivo. One of the signal transduction pathways

that has been associated with bone and cartilage formation,

but for which relatively little is known with relation to

Table 1

hMSC donor information

Donor Age Sex Source Passage Frozen/fresh

1 25 m iliac 2 frozen

2 66 f iliac 4 frozen

3 100 f iliac 2 frozen

4 72 f iliac 3 frozen

5 72 m iliac 2 frozen

6 77 f iliac 2 frozen

J. de Boer et al. / Bone 34 (2004) 818–826 819

hMSCs, is Wnt signaling. Wnts are a family of secreted

glycoproteins that initiate a signal transduction cascade upon

binding to the frizzled family of receptors and their low-

density lipoprotein-related protein (LRP) co-receptors (see

Refs. [4,34] and references therein). The Wnt signaling

pathway acts via the bipartite transcription factor h-cate-nin/T cell factor (TCF), which binds to the promoter of Wnt

responsive genes and thus initiates their transcription. In

unstimulated cells, cytoplasmic h-catenin is phosphorylated

by a complex of proteins containing axin, the adenomatous

polyposis coli protein (APC) and glycogen synthase kinase 3

(GSK3), which earmarks h-catenin for degradation by the

proteasome. Upon binding of Wnt to frizzled, the axin/APC/

GSK3 complex is inactivated, resulting in the accumulation

of cytoplasmic h-catenin, which will translocate to the

nucleus and activate Wnt responsive genes. The Wnt signal

transduction pathway has been implicated in bone formation:

patients suffering from osteoporosis–pseudoglioma syn-

drome have an inactivating mutation in the Wnt co-receptor

LRP5 [17], whereas an activating LRP5 mutation is associ-

ated with high bone mass syndrome [3,29]. Analysis of

LRP5-deficient mice revealed a decreased number of osteo-

blasts suggesting that Wnt signaling stimulates bone forma-

tion at the level of osteoprogenitor proliferation [24]. We

previously investigated the effect of Wnt over activation on

hMSCs and discovered that Wnt3A and low concentrations

(4 mM) of the Wnt mimic lithium stimulated hMSC prolif-

eration [10]. Furthermore, cells that were extensively ex-

panded in the presence of 4 mM lithium chloride could still

differentiate into both the osteogenic and adipogenic line-

ages. We concluded that both lithium and recombinant

Wnt3A might be used as mitogenic stimuli during hMSC

expansion in vitro or during bone tissue engineering in vivo.

In addition to its role in proliferation, some evidence sug-

gests that Wnt signaling can also stimulate osteogenesis.

Constitutive activation of Wnt signaling by retroviral trans-

fection of a stabilized form of h-catenin stimulated ALP

expression in C3H10T1/2 and ST2 osteogenic cells, whereas

wild type h-catenin failed to do so [1,17]. This suggests that

high levels of Wnt signaling can stimulate osteoblast differ-

entiation. This is further supported by the fact that ALP is

only induced at high concentrations of lithium in both

C3H10T1/2 cells [1] and hMSCs [10]. We want to exploit

the positive effect of Wnt signaling on both proliferation and

differentiation of osteogenic cells by controlled release of

lithium from porous calcium phosphate scaffolds and calci-

um phosphate coatings that we use for bone tissue engineer-

ing [11]. Because high lithium concentrations severely

inhibit hMSC proliferation [10], we aim at releasing low,

mitogenic levels of lithium. To assess the desired duration of

lithium release, we set to study how low levels of Wnt

signaling affect the osteogenic capacity of hMSCs. In this

paper, we report that Wnt signaling inhibits ALP expression

during dexamethasone-induced osteogenesis in hMSCs and

blocks mineralization of osteogenic hMSCs. Gene expres-

sion profiling of Wnt-stimulated hMSCs further supports the

conclusion that Wnt signaling inhibits rather than stimulates

osteogenic differentiation of hMSCs.

Materials and methods

Cell culturing

Bone marrow aspirates (5–15 ml) were obtained from six

donors that had given written informed consent. Donor

information is summarized in Table 1. hMSCs were isolated

and proliferated as described previously [11]. Briefly, aspi-

rates were resuspended using a 20 G needle, plated at a

density of 5 � 105 cells/cm2, and cultured in hMSC

proliferation medium, which contains minimal essential

medium (a-MEM, Life Technologies), 10% fetal bovine

serum (FBS, Life Technologies), 0.2 mM ascorbic acid

(Asap, Life Technologies), L-glutamin (Life Technologies),

100 U/ml penicillin (Life Technologies), 10 Ag/ml strepto-

mycin (Life Technologies), and 1 ng/ml basic fibroblast

growth factor (bFGF, Instruchemie, The Netherlands). Cells

were grown at 37jC in a humid atmosphere with 5% CO2.

Medium was refreshed twice a week and cells were used for

further subculturing or cryopreservation upon reaching near

confluence. hMSC basic medium was composed of hMSC

proliferative medium without bFGF, and hMSC osteogenic

medium was composed of hMSC basic medium supple-

mented with 10�8 M dexamethasone (Sigma) and 0.01 M

h-glycerol phosphate (Sigma). A lithium chloride (LiCl,

CalBiochem) stock solution of 400 mM in a-MEM and a

400 mM sodium chloride (NaCl, CalBiochem) solution in

phosphate-buffered saline solution (PBS) were used

throughout the study. L cells and L-Wnt3A cells were

obtained from the American Type Culture Collection

(CRL-2647 and CRL-2648, respectively). Control- and

Wnt3A-conditioned media were prepared essentially as

described by the supplier. Briefly, confluent cells were split

1:10, grown in basic hMSC medium, and conditioned

medium was collected after 4 and 7 days. Batches were

mixed and filter sterilized.

Flow cytometry

The effect of Wnt signaling on ALP expression was

studied by flow cytometry on cells seeded at 1000 cells/

J. de Boer et al. / Bone 34 (2004) 818–826820

cm2 in 6-well plates and grown under various conditions for

4 to 5 days. Each experiment was performed in triplicate and

included a negative control (cells grown in basic medium), a

positive control (cells grown in osteogenic medium), and one

or more experimental conditions (4 mM LiCl, different

concentrations of control- and Wnt3A-conditioned medium).

The effect of 4 mM LiCl on ALP levels was analyzed in cells

of donors 2, 3, and 4, the effect of Wnt3Awas studied in cells

of donors 3, 4, and 5. After 4 or 5 days of culture, cells were

trypsinized and incubated for 30 min in PBS/5% bovine

serum albumin (BSA, Sigma) and incubated in PBS/1%

BSA plus primary antibody (anti-ALP B4-78 [Developmen-

tal Studies Hybridoma Bank, University of Iowa, USA]) for

30 min, washed three times in PBS/1% BSA, and incubated

with secondary antibody (goat anti-mouse IgG PE, DAKO)

for 30 min. Mouse IgG2a (DAKO) was used as an isotype

control. Cells were washed three times and suspended in 250

Al PBS/1% BSA plus 10 Al Viaprobe (PharMingen) for live/

dead cell staining. Staining was analyzed on a FACS Calibur

(Becton Dickinson Immunocytometry Systems) and ALP

levels were analyzed on living cells only, with a minimum of

7500 gated events.

Mineralization

For mineralization, hMSCs were seeded in duplo at 1000

cells/cm2 in T25 culture flasks and grown under various

conditions. In every experiment, osteogenic medium was

used as a positive control and basic medium as negative

control. Cells were rinsed with PBS and fixed overnight in

4% paraformaldehyde (Merck) when extensive mineraliza-

tion was observed in the positive control by phase contrast

microscopy, which usually occurred between 21 and 28

days of culture. Next, the cells were rinsed with deminer-

alized water and incubated in 5% silver nitrate (Sigma) until

a distinct black stain was observed in the positive control.

The mineralization experiment was performed with cells of

donors 3, 5, and 6. Mineralization was quantified by image

analysis of the total T25 area. Images were taken using a

Sony Mavica model MUV-FD85 digital camera and miner-

alized area was expressed as percentage of total area using

ImageJ imaging software (http://rsb.info.nih.gov/ij/).

Micro-array analysis

To assess the effect of Wnt pathway activation at the

transcriptional level, hMSCs of donor 1 were grown for 4

days in basic medium, basic medium supplemented with 4

mM LiCl, basic medium supplemented with 10% control-

conditioned medium, and basic medium supplemented with

10% Wnt3A-conditioned medium. RNA was isolated using

a RNeasy midi kit (Qiagen) and 30 Ag of total RNA was

used for probe labeling according to the manufacturer’s

protocol (Affymetrix). Probe quality was verified using lab-

on-chip technology (Agilent Technologies) and samples

were hybridized to Human Genome Focus arrays according

to manufacturer’s protocol (Affymetrix). Data analysis was

performed using Affymetrix GENECHIP 4.0 software.

Quantitative PCR

The effect of LiCl and Wnt3A on S100A4 mRNA levels

was determined by seeding hMSCs of donors 3, 4, and 5 at

5000 cells/cm2 in T25 culture flasks in 5 ml of basic

medium, basic medium supplemented with 4 mM LiCl,

basic medium supplemented with 10% control-conditioned

medium, and basic medium supplemented with 10%

Wnt3A-conditioned medium. Total RNA was isolated using

an RNeasy mini kit (Qiagen) and on column DNase treated

with 10U RNase-free DNase I (Gibco) at 37jC for 30 min.

DNase was inactivated at 72jC for 15 min. The quality and

quantity of RNA was analyzed by gel electrophoresis and

spectrophotometry. Two micrograms of each DNase-treated

RNA sample was used for first strand cDNA synthesis using

Superscript II (Invitrogen) according to the manufacturers’

protocol. One microliter of 100� diluted cDNA was used

for 18s rRNA control amplification, and 1 Al of undilutedcDNA was used for S100A4 amplification. PCR reactions

were performed and monitored on a Light Cycler real time

PCR machine (Roche) using the SYBR Green I master mix

(Eurogentec) with primers for 18s rRNA (18srRNA-F

5Vcggctaccacatccaaggaa3V and 18srRNA-R 5Vgctggaatta-ccgcgggt3V) and for S100A4 (S100A4-F 5Vagcttcttgg-ggaaaaggac3Vand S100A4-R 5Vccccaaccacatcaa-gagg3V).Data was analyzed using Light Cycler software version

3.5.3, using the fit point method by setting the noise band

to one. Expression of S100A4 was calculated relative to 18s

rRNA levels by comparative DCT method [30]. Each sample

was analyzed at least in duplicate and averages were used

for further calculations.

IL-6 ELISA

The effect of LiCl and Wnt3A on IL-6 secretion by

hMSCs was determined by seeding hMSCs of donors 3

and 4 in triplicate at 5000 cells/cm2 in T25 culture flasks in 5

ml of basic medium, basic medium supplemented with 4 mM

LiCl, basic medium supplemented with 10% control-condi-

tioned medium, and basic medium supplemented with 10%

Wnt3A-conditioned medium. Conditioned media were col-

lected from the hMSC cultures after 4 days and IL-6 levels

were determined using a human IL-6 ELISA kit (Pierce)

according to the manufacturers’ protocol.

Results

Wnt signaling inhibits ALP expression in differentiating

hMSCs

In previous studies, we noticed that conditioned medium

from mouse L cells, which we used as control-conditioned

J. de Boer et al. / Bone 34 (2004) 818–826 821

medium (see Materials and Methods), induces the expression

of the osteogenic marker alkaline phosphatase (ALP) in

hMSCs (see Ref. [10] and compare ALP levels among the

negative control, 10 c, and 50 c in Fig. 1). Interestingly,

conditioned medium from Wnt3A transgenic L cells

(Wnt3A-conditioned medium) did not show an increase in

ALP expression. Therefore, we hypothesized that Wnt

signaling could inhibit ALP expression in hMSCs. To test

this hypothesis, we grew hMSCs in osteogenic medium,

which contains dexamethasone, and analyzed ALP expres-

sion after 5 days. As expected, dexamethasone stimulated

ALP expression in hMSCs (see Fig. 1). To induce Wnt

signaling during osteogenic differentiation, we added lithium

chloride, which inhibits the negative Wnt signaling regulator

GSK3 and thus stimulates Wnt signaling [26]. As depicted in

Fig. 1, 4 mM LiCl reduced ALP induction in osteogenic

medium from 328 F 30% to 206 F 17% of ALP expression

in control medium. Next, we stimulated Wnt signaling by

supplementing osteogenic medium with 10% and 50%

Wnt3A- or control-conditioned medium, respectively. The

strongest effect was seen with 50% Wnt3A-conditioned

medium, which inhibited ALP expression to levels compa-

rable to those observed in unstimulated hMSCs (Fig. 1).

Wnt signaling inhibits mineralization of hMSCs

Inhibition of ALP by lithium and Wnt3A in hMSCs

undergoing osteogenesis suggests that Wnt signaling might

Fig. 1. Lithium chloride and Wnt3A inhibit dex-induced ALP expression in

hMSCs. ALP levels determined by flow cytometry in hMSCs of donor 3,

grown in different culture conditions: �, basic medium; +, osteogenic

medium; Li, osteogenic medium supplemented with 4 mM LiCl; 10 c,

osteogenic medium supplemented with 10% control-conditioned medium;

10 w, osteogenic medium supplemented with 10% Wnt3A-conditioned

medium; 50 c, osteogenic medium supplemented with 50% control-

conditioned medium; 50 w, osteogenic medium supplemented with 50%

Wnt3A-conditioned medium. ALP is expressed relative to the level in the

negative control. Error bars represent standard deviation of triplicate

experiments. Data was analyzed by Student’s t test and ALP levels differed

significantly between + and Li, 10 c and 10w, and 50 c and 50w, respectively

( P < 0.01).

block dexamethasone-induced osteogenesis. We therefore

investigated the effect of Wnt signaling on the endpoint of

osteogenesis, that is, mineralization. In Fig. 2A, we show

that hMSCs grown in the presence of dexamethasone and h-glycerol phosphate display extensive mineralization whereas

cells grown in the absence of dexamethasone do not. We then

studied how mineralization is affected upon stimulation of

Wnt signaling. As shown in Fig. 2A and Table 2, 10%

Wnt3A-conditioned medium, but not control-conditioned

medium, completely blocked dexamethasone-induced min-

eralization of hMSCs in cells from all three donors studied.

Similarly, mineralization in lithium-treated cultures of

donors 3 and 5 was strongly reduced compared to cells

grown in osteogenic medium although some mineralization

could still be observed in cells from donor 5. Surprisingly,

lithium had a stimulatory effect on mineralization in cells of

donor 6 (data not shown). As a control, mineralization in

hMSC grown in osteogenic medium supplemented with 4

mM NaCl was not noticeably affected in cells from any

donor studied.

To delineate the developmental window in which Wnt

signaling can block hMSC mineralization, we allowed the

cells to grow in osteogenic medium. Examination of the

cultures showed that the first signs of mineralization

appeared after 12 days of culture, with small areas of

mineral deposition (data not shown). After 17 days, the

culture was fully mineralized with large areas of minerali-

zation covering the cell sheet (see Fig. 2B). In parallel, we

allowed the cells to grow in osteogenic medium for 0, 4, 8,

and 12 days before adding 10% Wnt3A- or control-condi-

tioned medium and determined the effect on mineralization

after 17 days. As expected, Wnt3A exposure from day 0

onwards completely blocked mineralization. Similarly, min-

eralization was inhibited when Wnt3A was added at days 4

and 8. In contrast, when Wnt3A was added at day 12, we

observed a fully mineralized cell sheet at day 17. Control-

conditioned medium did not have an effect on mineraliza-

tion (data not shown).

To examine whether Wnt3A affects extracellular matrix

deposition, hMSCs were grown on titanium discs for 21

days and analyzed for matrix formation by scanning elec-

tron microscopy. As expected, both hMSCs grown in basic

medium and hMSCs grown in osteogenic medium depos-

ited an extensive extracellular matrix (Figs. 2C, D). Al-

though we did not observe gross abnormalities in

morphology of the collagen fibers deposited by cells grown

in osteogenic medium supplemented with Wnt3A-condi-

tioned medium, the matrix did appear slightly disorganized

(Fig. 2E).

Micro-array analysis of hMSCs treated with lithium and

Wnt3A

We previously discovered that Wnt signaling stimulates

proliferation of hMSCs but does not noticeably affect hMSC

pluripotency [10]. To acquire a global view of the cellular

Fig. 2. Lithium chloride and Wnt3A inhibit dex-induced mineralization of hMSCs. (A) Mineralization in hMSCs of donor 5 grown under various conditions for

27 days: �, basic medium; +, osteogenic medium; Li, osteogenic medium supplemented with 4 mM LiCl; Na, osteogenic medium supplemented with 4 mM

NaCl; 10 c, osteogenic medium supplemented with 10% control-conditioned medium; 10 w, osteogenic medium supplemented with 10% Wnt3A-conditioned

medium. Mineralization was visualized by von Kossa staining. The scale bar is 2 mm. (B) Mineralization in hMSCs of donor 5 grown for 17 days in basic

medium (�), osteogenic medium (+), or in osteogenic medium supplemented with 10% Wnt3A-conditioned medium at different time points after the start of

the experiment (+w d0, Wnt3A added at day 0; etc.). Mineralization was visualized by von Kossa staining. The scale bar is 2 mm. (C, D, E) Scanning electron

microscopical images of hMSCs of donor 5 grown for 21 days in basic medium (C), osteogenic medium (D), or osteogenic medium supplemented with 10%

Wnt3A-conditioned medium (E). Note the fibers that indicate extensive matrix deposition. The scale bar is 1 Am.

J. de Boer et al. / Bone 34 (2004) 818–826822

processes affected by Wnt activation, we decided to study

gene expression profiles of hMSCs grown for 4 days in

basic medium, basic medium supplemented with 4 mM

lithium chloride, with 10% control-conditioned medium,

and with 10% Wnt3A-conditioned medium. RNA was

isolated and gene expression was analyzed on Affymetrix

Human Genome Focus arrays. Gene expression was com-

Table 2

Mineralization in osteogenic hMSCsa

Donor

3bDonor

5

Donor

3

Donor

5

Donor

6

�dex 1 0 0 0 0 2 �dex 1

�dex 2 0 0 0 0 1 �dex 2

dex 1 45 13 21 12 16 dex + control 1

dex 2 41 20 16 18 17 dex + control 2

dex + li1 0 4 3 0 2 dex + wnt3A 1

dex + li2 0 3 1 0 6 dex + wnt3A 2

a Mineralisation in osteogenic cultures, indicated as percentage mineral-

ized area of total area. Duplicate experiments are shown. �dex, basic

medium; dex, osteogenic medium; dex + li, osteogenic medium plus 4

mM lithium; dex + control, osteogenic medium plus 10% control-

conditioned medium; dex + wnt3A, osteogenic medium plus 10% wnt3A-

conditioned medium.b Every column represents an independent experiment, for donor numbers,

refer to Table 1.

pared between cells grown in control- to Wnt3A-condi-

tioned medium, and in basic versus lithium chloride-

supplemented medium. We then focused on the genes

similarly regulated for at least 1.3-fold by both lithium

and Wnt3A (summarized in Table 3). Out of the approxi-

mately 9000 transcripts on the array, only 37 genes matched

these criteria and only 4 genes were regulated more than

twofold in both conditions analyzed (see Table 3). Of the 37

differentially regulated genes, 32 were downregulated and 5

genes were upregulated when compared to nonsupple-

mented medium. IL-6 was one of the most strongly regu-

lated genes on the array, and to validate our micro-array

data, we analyzed IL-6 secretion by hMSCs of two different

donors, grown in basic medium, basic medium supple-

mented with lithium, or basic medium supplemented with

either control- or Wnt3A-conditioned medium for 4 days.

As shown in Fig. 3A, both Wnt3A and lithium significantly

inhibit IL-6 secretion by hMSCs. To further confirm our

micro-array data, we isolated RNA from the cells described

in the previous experiment and analyzed S100A4 expression

using quantitative PCR. As expected, S100A4 RNA levels

were upregulated by both Wnt3A- and lithium-exposed cells

(Fig. 3B).

Lithium- and Wnt3A-treated hMSCs display a distinct

increase in cell proliferation, and accordingly, four genes

Fig. 3. Wnt signaling regulates IL-6 and S100A4 expression in hMSCs. (A)

IL-6 concentration in medium conditioned for 4 days by hMSCs of donor 4.

hMSCs were grown in basic medium (�), basic medium supplemented with

4 mM LiCl (Li), basic medium supplemented with 10% control-conditioned

medium (10 c), or basic medium supplemented with 10% Wnt3A-

conditioned medium (10 w). IL-6 levels were determined by ELISA. Each

experiment was performed in triplicate and error bars indicate the standard

deviation. Data was analyzed using Student’s t test and a statistically

significant difference was found between � and Li, and 10 c and 10 w,

respectively ( P < 0.01). (B) S100A4 mRNA expression in hMSCs

determined by quantitative PCR. Cells were grown in basic medium (�),

and basic medium supplemented with either 4 mM lithium (Li), 10%

control-conditioned medium (control), or 10% Wnt3A-conditioned medium

(Wnt). Values represent the average S100A4 expression level in cells from

three different donors. The bar indicates the standard error of mean.

S100A4 level in lithium-treated cells is expressed relative to the level in

cells grown in basic medium; similarly S100A4 levels in Wnt3A-treated

cells are normalized to control-treated cells. A statistical significant

difference was found between � and Li and between 10 c and 10 w using

Student’s t test ( P < 0.05).

Table 3

Genes regulated by both Wnt3A and lithium chloride in hMSCs

Osteogenesis

ENPP1 (�1.5/�1.3)a

ID2 (�1.7/�1.4)

Transglutaminase (�1.9/�1.4)

VitD3 upregulated (�1.5/�1.4)

Leptin receptor (�1.6/�2.1)

S100A4 (1.9/1.3)

Proenkephalin (�1.3/�3.2)

Hematopoiesis

SDF-1 (�1.9/�1.7)

SLIT-2 (1.3/1.4)

EPAS-1 (�1.3/�1.3)

IL-6 (�2.3/�2.5)

Chondrogenesis

GDF-5 (�1.6/�1.7)

Collagen X (�1.6/�1.9)

Sox-4 (�1.7/�1.3)

Proliferation

meox-2 (�2.3/�1.6)

cpr8 (�1.4/�1.5)

btg-1 (�1.4/�1.3)

CREG (�1.4/�1.3)

Other mesenchymal genes

EFEMP-1 (�2.1/�2.5)

Dystrophin (�1.9/�2.5)

Laminin a4 (�1.4/�1.3)

SM22 (1.4/1.4)

dsc54 (�1.5/�1.5)

Miscellaneous

PTX3 (�2.3/�3.0)

Cytochrome P450 (�1.5/�1.5)

Phosphatidic acid phosphatase 2A (�1.9/�1.4)

HLA-DMA (�1.4/�1.5)

MHC gamma chain (�1.4/�1.5)

Keratin 14 (1.6/1.4)

TEM7 (2.3/2.1)

Cited-2/MRG1 (�1.3/�1.6)

Dihydropyrimidine dehydrogenase (�1.5/�1.3)

Myomegalin (�1.9/�1.3)

PKC A (�1.3/�1.3)

Desmoplakin (�1.5/�1.4)

Histamin N-methyltransferase (�1.4/�1.3)

lmcd-1 (�1.5/�1.6)

a Fold regulation in control- versus Wnt3A-treated cells and non- versus

lithium-treated cells is indicated between parentheses (left and right,

respectively).

J. de Boer et al. / Bone 34 (2004) 818–826 823

involved in proliferation were differentially regulated. Mes-

enchymal-specific homeobox gene 2 (MEOX-2), btg-1 [38]

and CREG [44] are negative regulators of proliferation

[19,41] and were downregulated. However, a presumed

positive regulator of the cell cycle [13], cell cycle progression

8 protein, was also downregulated. Notably, undifferentiated

hMSCs express many osteogenesis- or chondrogenesis-spe-

cific genes (e.g., ID-2 [21,27,31], transglutaminase [23],

collagen type X and Sox-4 [39], most of which are down-

regulated by bothWnt3A and lithium. This suggests that Wnt

signaling inhibits the differentiated phenotype of hMSCs,

which is further supported by the fact that the negative

J. de Boer et al. / Bone 3824

regulator of mineralization S100A4 is upregulated by both

Wnt3A and lithium.

Discussion

Wnt signaling inhibits osteogenic differentiation of hMSCs

Dexamethasone-induced osteogenesis in hMSCs is char-

acterized by an increase in ALP expression after 4 days,

followed by matrix deposition, matrix maturation, and

mineralization at later stages [8,22,32]. Bone-specific ALP

is a member of a family of three ALP proteins required for

phosphate homeostasis [14,35], and inhibition is expected to

affect mineralization. Although Wnt overactivation did not

overtly affect matrix deposition, ALP expression was

inhibited. Moreover, micro-array analysis revealed that

Wnt signaling downregulates transglutaminase, which

cross-links proteins in the extracellular matrix. Downregu-

lation of these proteins suggests that Wnt signaling inhibits

expression of genes directly involved in the formation of a

mineralized bone matrix. Interestingly, micro-array analysis

further revealed that lithium and Wnt3A downregulate ID2

gene expression. ID2 is a direct target gene of the BMP

signal transduction cascade [21,27,31], which suggests that

Wnt signaling can also interfere with early events of

osteogenic differentiation. Because we performed the mi-

cro-array experiment after 4 days of exposure to Wnt3A or

lithium, the effect of Wnt signaling on ID2 expression might

be secondary. Future research has to delineate the direct

targets of Wnt signaling and the subsequent events that lead

to inhibition of osteogenesis.

Wnts are secreted proteins that act as morphogens; that

is, a protein can elicit differential responses in the same cell

type depending on its concentration [4]. In this study, we

show that low levels of lithium and Wnt3A inhibit dex-

induced osteogenesis in hMSCs, whereas previous studies

have shown that high levels of Wnt signaling can induce

ALP activity. For instance, we have shown that 40 mM, but

not 4 mM lithium, induces ALP expression in hMSCs [10],

and Bain et al. [1] showed similar results in C3H10T1/2

cells. Moreover, Gong et al. [17] show that a stabilized, and

therefore more active form of the Wnt effector molecule h-catenin, can induce ALP in C3H10T1/2 cells, whereas wild

type h-catenin failed to do so. Based on these data, we

propose that Wnt signaling has a dose-dependent effect on

osteogenic differentiation. At low levels, Wnt signaling

blocks differentiation and stimulates osteoprogenitor prolif-

eration, whereas at high levels, Wnt signaling stimulates

osteogenic differentiation. To further test dose-dependent

effects of Wnt signaling on osteogenesis, experiments have

to be designed in which the level of Wnt stimulation is

carefully controlled, for instance by using defined concen-

trations of recombinant Wnt3A or in mouse models with

defined molecular defects in the Wnt signaling pathway

[25].

Wnt downregulates differentiation-specific markers in

proliferating hMSCs

Our micro-array experiments demonstrate that proliferat-

ing, undifferentiated hMSCs express many cell lineage-

specific transcripts. Previously, Tremain et al. [43] elegantly

showed expression of multiple transcripts specific for bone,

cartilage, and hematopoiesis supporting stromal cells as well

as nonmesenchymal cell types such as neurons and endo-

thelial cells in a single undifferentiated hMSC colony. In our

analysis of genes that were regulated by both lithium and

Wnt3A, we observed downregulation of many genes that

are typical for differentiated cells such as transglutaminase

(osteogenesis), collagen type X (chondrogenesis), and stro-

mal cell-derived factor (SDF-1, hematopoiesis supporting

stromal cells [28]). In all cases, the lineage-specific genes

were downregulated. Interestingly, genes that negatively

regulated either osteogenesis (S100A4 [12]) or the hemato-

poiesis supporting function (SLIT2 [15]) were upregulated.

S100A4 is an intracellular calcium-binding protein ex-

pressed by osteoblastic cells. S100A4 is expressed during

early stages of MC3T3 osteogenesis and inhibits minerali-

zation. S100A4 is upregulated by Wnt3A and lithium in our

experiments, which suggests that Wnt signaling brings

hMSCs into a more primitive, less differentiated state.

Similarly, SDF-1 is a cytokine involved in chemotaxis of

hematopoietic cells and a marker for the hematopoiesis

supporting function of bone marrow stromal cells [28].

Both Wnt3A and lithium downregulate SDF-1, whereas

SLIT-2, a secreted antagonist of SDF-1 [15], is upregulated.

This demonstrates that exposing undifferentiated hMSCs to

Wnt signaling results in inhibition of the hematopoiesis

supporting function of hMSCs as described before [45]. An

exception to the rule that Wnt signaling downregulates

hMSC differentiation is the upregulation of SM22, a smooth

muscle specific gene [42], by both lithium and Wnt3A

(Table 3), suggesting that Wnt signaling induces smooth

muscle cell differentiation in hMSCs. This is consistent with

SM22 upregulation by Wnt signaling as reported in

C3H10T1/2 cells [1] and the role that Wnt signaling plays

in smooth muscle cell development [9,40]. Furthermore, the

profile of inhibition and stimulation of mesenchymal differ-

entiation that emerges from our study is strikingly similar to

the differentiation profile of embryonic stem cells in which

Wnt signaling is constitutively active [25]. We previously

studied mouse embryonic stem cells in which Wnt signaling

was constitutively activated by targeted inactivation of the

APC gene. Whereas wild-type ES cells form tissue of all

different germ layers when injected subcutaneously into

immune-deficient mice, APC mutant ES cells fail to differ-

entiate into the osteogenic and chondrogenic lineage where-

as abundant smooth muscle differentiation was observed

[25].

In conclusion, Wnt signaling inhibits dexamethasone-

induced osteogenesis in hMSCs, and in future bone tissue

engineering studies, we will try to limit release of lithium or

4 (2004) 818–826

J. de Boer et al. / Bone 34 (2004) 818–826 825

Wnt3A from calcium phosphate scaffolds to the proliferative

phase of osteogenesis.

Acknowledgment

The research of C.G. was sponsored by a grant from the

Association for International Cancer Research.

References

[1] Bain G, Muller T, Wang X, Papkoff J. Activated beta-catenin induces

osteoblast differentiation of C3H10T1/2 cells and participates in

BMP2 mediated signal transduction. Biochem Biophys Res Commun

2003;301:84–91.

[2] Bianco P, Robey PG. Stem cells in tissue engineering. Nature

2001;414:118–21.

[3] Boyden LM, Mao J, Belsky J, Mitzner L, Farhi A, Mitnick MA, et al.

High bone density due to a mutation in LDL-receptor-related protein

5. N Engl J Med 2002;346:1513–21.

[4] Cadigan KM, Nusse R. Wnt signaling: a common theme in animal

development. Genes Dev 1997;11:3286–305.

[5] Caplan AI, Bruder SP. Mesenchymal stem cells: building blocks for

molecular medicine in the 21st century. Trends Mol Med 2001;7:

259–64.

[6] Castro-Malaspina H, Gay RE, Resnick G, Kapoor N, Meyers P, Chia-

rieri D, et al. Characterisation of human bone marrow fibroblast colo-

ny-forming cells (CFU-F) and their progeny. Blood 1980;56:289–301.

[7] Caterson EJ, Nesti LJ, Li WJ, Danielson KG, Albert TJ, Vaccaro AR,

et al. Three-dimensional cartilage formation by bone marrow-derived

cells seeded in polylactide/alginate amalgam. J Biomed Mater Res

2001;57:394–403.

[8] Cheng SL, Yang JW, Rifas L, Zhang SF, Avioli LV. Differentiation of

human bone marrow osteogenic stromal cells in vitro: induction of the

osteoblast phenotype by dexamethasone. Endocrinology 1994;134:

277–86.

[9] Cossu G, Borello U. Wnt signaling and the activation of myogenesis

in mammals. EMBO J 1999;18:6867–72.

[10] de Boer J, Wang HJ, van Blitterswijk CA. Effects of Wnt signalling

on proliferation and differentiation of human mesenchymal stem cells.

Tissue Eng 2004;10:393–401.

[11] de Bruijn JD, van den Brink I, Mendes S, Dekker R, Bovell YP, van

Blitterswijk CA. Bone induction by implants coated with cultured

osteogenic bone marrow cells. Adv Dent Res 1999;13:74–81.

[12] Duarte WR, Shibata T, Takenaga K, Takahashi E, Kubota K, Ohya K,

et al. S100A4: a novel negative regulator of mineralization and oste-

oblast differentiation. J Bone Miner Res 2003;18:493–501.

[13] Edwards MC, Liegeois N, Horecka J, DePinho RA, Sprague Jr GF,

Tyers M, et al. Human CPR (cell cycle progression restoration) genes

impart a Far-phenotype on yeast cells. Genetics 1997;147:1063–76.

[14] Fedde KN, Blair L, Silverstein J, Coburn SP, Ryan LM, Weinstein

RS, et al. Alkaline phosphatase knock-out mice recapitulate the met-

abolic and skeletal defects of infantile hypophosphatasia. J Bone

Miner Res 1999;14:2015–26.

[15] Fernandis AZ, Ganju RK. Slit: a roadblock for chemotaxis. Sci Signal

Transduct Knowl Environ 2001;2001:PE1.

[16] Friedenstein AJ, Piatetzky-Shapiro IL, Petrakova KV. Osteogenesis in

transplants of bone marrow cells. J Embryol Exp Morphol 1966;16:

381–90.

[17] Gong Y, Slee RB, Fukai N, Rawadi G, Roman-Roman S, Reginato

AM, et al. LDL receptor-related protein 5 (LRP5) affects bone accrual

and eye development. Cell 2001;107:513–23.

[18] Goshima J, Goldberg VM, Caplan AI. The osteogenic potential of

culture-expanded rat marrow mesenchymal cells assayed in vivo in

calcium phosphate ceramic blocks. Clin Orthop 1991;298–311.

[19] Grigoriou M, Kastrinaki MC, Modi WS, Theodorakis K, Mankoo B,

Pachnis V, et al. Isolation of the human MOX2 homeobox gene and

localization to chromosome 7p22.1 –p21.3. Genomics 1995;26:

550–5.

[20] Haynesworth SE, Goshima J, Goldberg VM, Caplan AI. Character-

ization of cells with osteogenic potential from human marrow. Bone

1992;13:81–8.

[21] Hollnagel A, Oehlmann V, Heymer J, Ruther U, Nordheim A. Id

genes are direct targets of bone morphogenetic protein induction in

embryonic stem cells. J Biol Chem 1999;274:19838–45.

[22] Jaiswal N, Haynesworth SE, Caplan AI, Bruder SP. Osteogenic dif-

ferentiation of purified, culture-expanded human mesenchymal stem

cells in vitro. J Cell Biochem 1997;64:295–312.

[23] Kaartinen MT, El-Maadawy S, Rasanen NH, McKee MD. Tissue

transglutaminase and its substrates in bone. J Bone Miner Res 2002;

17:2161–73.

[24] Kato M, Patel MS, Levasseur R, Lobov I, Chang BH, Glass II DA,

et al. Cbfa1-independent decrease in osteoblast proliferation, osteo-

penia, and persistent embryonic eye vascularization in mice defi-

cient in Lrp5, a Wnt coreceptor. J Cell Biol 2002;157:303–14.

[25] Kielman MF, Rindapaa M, Gaspar C, van Poppel N, Breukel C, van

Leeuwen S, et al. Apc modulates embryonic stem-cell differentiation

by controlling the dosage of beta-catenin signaling. Nat Genet

2002;32:594–605.

[26] Klein PS, Melton DA. A molecular mechanism for the effect of

lithium on development. Proc Natl Acad Sci U S A 1996;93:8455–9.

[27] Korchynskyi O, Dechering KJ, Sijbers AM, Olijve W, ten Dijke P.

Gene array analysis of bone morphogenetic protein type I receptor-

induced osteoblast differentiation. J Bone Miner Res 2003;18:

1177–85.

[28] Lapidot T, Petit I. Current understanding of stem cell mobilization: the

roles of chemokines, proteolytic enzymes, adhesion molecules, cyto-

kines, and stromal cells. Exp Hematol 2002;30:973–81.

[29] Little RD, Carulli JP, Del Mastro RG, Dupuis J, Osborne M, Folz C,

et al. A mutation in the LDL receptor-related protein 5 gene results in

the autosomal dominant high-bone-mass trait. Am J Hum Genet

2002;70:11–9.

[30] Livak KJ, Schmittgen TD. Analysis of relative gene expression data

using real-time quantitative PCR and the 2(-Delta Delta C(T)) Meth-

od. Methods 2001;25:402–8.

[31] Locklin RM, Riggs BL, Hicok KC, Horton HF, Byrne MC, Khosla S.

Assessment of gene regulation by bone morphogenetic protein 2 in

human marrow stromal cells using gene array technology. J Bone

Miner Res 2001;16:2192–204.

[32] Maniatopoulos C, Sodek J, Melcher AH. Bone formation in vitro by

stromal cells obtained from bone marrow of young adult rats. Cell

Tissue Res 1988;254:317–30.

[33] Mendes SC, Tibbe JM, Veenhof M, Bakker K, Both S, Platenburg PP,

et al. Bone– tissue engineered implants using human bone marrow

stromal cells: effect of culture conditions and donor age. Tissue Eng

2002;8:911–20.

[34] Moon RT, Bowerman B, Boutros M, Perrimon N. The promise and

perils of Wnt signalling through beta-catenin. Science 2002;296:

1644–6.

[35] Narisawa S, Frohlander N, Millan JL. Inactivation of two mouse

alkaline phosphatase genes and establishment of a model of infantile

hypophosphatasia. Dev Dyn 1997;208:432–46.

[36] Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca

JD, et al. Multilineage potential of adult human mesenchymal stem

cells. Science 1999;284:143–7.

[37] Pittenger MF, Marshak DR. Adult mesenchymal stem cells. In: Got-

tlieb D, editor. Stem Cell Biology, vol. 1. Cold Spring Harbor, New

York: Cold Spring Harbor Laboratory Press; 2001. p. 349–73.

[38] Prevot D, Morel AP, Voeltzel T, Rostan MC, Rimokh R, Magaud JP,

et al. Relationships of the antiproliferative proteins BTG1 and BTG2

J. de Boer et al. / Bone 34 (2004) 818–826826

with CAF1, the human homolog of a component of the yeast CCR4

transcriptional complex: involvement in estrogen receptor alpha sig-

naling pathway. J Biol Chem 2001;276:9640–8.

[39] Reppe S, Rian E, Jemtland R, Olstad OK, Gautvik VT, Gautvik KM.

Sox-4 messenger RNA is expressed in the embryonic growth plate

and regulated via the parathyroid hormone/parathyroid hormone-

related protein receptor in osteoblast-like cells. J Bone Miner Res

2000;15:2402–12.

[40] Shi DL, Bourdelas A, Umbhauer M, Boucaut JC. Zygotic Wnt/beta-

catenin signaling preferentially regulates the expression of Myf5 gene

in the mesoderm of Xenopus. Dev Biol 2002;245:124–35.

[41] Smith RC, Branellec D, Gorski DH, Guo K, Perlman H, Dedieu JF,

et al. p21CIP1-mediated inhibition of cell proliferation by overexpres-

sion of the gax homeodomain gene. Genes Dev 1997;11:1674–89.

[42] Sobue K, Hayashi K, Nishida W. Molecular mechanism of phenotyp-

ic modulation of smooth muscle cells. Horm Res 1998;50(Suppl 2):

15–24.

[43] Tremain N, Korkko J, Ibberson D, Kopen GC, DiGirolamo C,

Phinney DG. MicroSAGE analysis of 2353 expressed genes in a

single cell-derived colony of undifferentiated human mesenchymal

stem cells reveals mRNAs of multiple cell lineages. Stem Cells

2001;19:408–18.

[44] Veal E, Groisman R, Eisenstein M, Gill G. The secreted glycoprotein

CREG enhances differentiation of NTERA-2 human embryonal car-

cinoma cells. Oncogene 2000;19:2120–8.

[45] Yamane T, Kunisada T, Tsukamoto H, Yamazaki H, Niwa H, Takada

S, et al. Wnt signaling regulates hemopoiesis through stromal cells.

J Immunol 2001;167:765–72.